In one of the most comprehensive reports from Anthropic’s threat intelligence team yet, the Claude-maker has said that multiple state-backed actors used its AI model to do weapons research, including missile navigation systems, kamikaze drone swarms and biological weapons.
These users mostly originated from usual geographical suspects in such cases — Iran, China, Russia and Yemen. However, none of the cases identified by Anthropic involved its recently released advanced models — Fable & Mythos.
“We disrupted every operation in the report, and used the lessons from them to strengthen our safeguards. Where appropriate, we also shared what we found with authorities and other AI companies”, Anthropic said in a statement.
The company also highlighted, that most of these highlighted cases are not typical: these turned out to be some of the most sophisticated misuse reported thus far.
Between December 2025 and August 2026, Anthropic’s threat intelligence team identified and disrupted operations across seven harm categories: cyber operations, influence operations, surveillance, scams and fraud, biological misuse, conventional weapons development, and unauthorised model replication.
Three conventional weapons cases involved China-linked actors, two involved Russia, and one was attributed to actors affiliated with Ansar Allah — the Houthi movement that has sustained strikes against energy infrastructure and commercial shipping throughout 2026. The attempted applications included kamikaze drone swarms and missile navigation systems.
All misuse cases involved Claude Haiku, Sonnet, and Opus. None involved Claude Fable or Mythos, with the exception of one distillation attempt — where a bad actor attempted to replicate Claude’s capabilities into an unrestricted model.
Even more alarmingly, Anthropic flagged five cases in which researchers used Claude in ways that could have supported biological weapons development. In one instance from May 2026, a scientist sought help drafting a grant application for gain-of-function research on the chikungunya virus — a mosquito-borne pathogen with no licensed treatment — with the proposal aimed at increasing transmissibility and immune evasion. The research was affiliated with a military research institute. The company also flagged separate gain-of-function work involving bird flu.